... phase; DA, double antibody; PEG, polyethyleneglycol 6000; GB. glass beads; PB, polystyrene beads; IRMA, immunoradiometric assay. ..... Thomas G. Rosano.
CLIN. CHEM. 30/2, 298-301 (1984)
Use of a Reference Standard to Improve the Accuracy and Precision of Seven Kitsfor Determination of Ferritin in Serum Carmelo lacobeilo,
Salvatore Ghlelmi,’ Slivana BeiIoII,1Paolo Arosio,2and Alberto Albertinl1
We assessed the accuracy and precision of seven commercial kits for serum ferritin. The concentration of ferritin as determined
by these
assays
for a liver ferritin
reference
standard, a human heart ferritin standard, and normal sera correlated highly, but the absolute values varied widely.Use of an international reference standard for ferritin to prepare the standard curve greatlydiminishedthe variability, butdid not eliminate it. Although the seven kits differed in specificity for various human isolerritins, this did not appear substantially to affect accuracy. Six of the seven kits appeared sufficiently precise for clinical use over a wide range of ferritin concentrations.
also studied the influence of each kit.
of serum matrix
and the precision
Materials and Methods Reference human liver standard (no. 80/602) was obtained from the Medical Research Council, National Biological Standard Board, Hampstead-Holly Hill, London (hereafter called the “NBSB standard”). Human heart ferritin was purified as previously described (11). Sera from normal subjects, mixed to obtain three pools with different concen-
trations of serum ferritin, were stored frozen until used. Commercial serum ferritin kits were obtained from Lopetit, Milano, Italy;ClinicalAssays, Division of Travenol Laboratories Inc., Cambridge, MA; Leeco Diagnostic Inc., Addftional K.yphrases: isoferritins . analytical error referSouthfleld, MI; Corning Medical and Scientific Laboraence materials tories, Medfleld, MA; Behringwerke AG, Marburg, F.R.G.; Ramco Laboratories Inc., Houston, TX; and Hybritech Inc., The concentration of ferritin in serum ordinarily reflects San Diego, CA. Table 1 summarizes their major technical body iron stores (1) and is considered one of the most reliable indicators of the iron status of patients (2,3). The relatively characteristics. We closely followed the instructions enclosed in the packages, and for all dilutions of sara and low concentration of ferritin in serum can be detected only standards we used the respective buffers providedin the with sensitive immunological methods such as radio- and kits. enzyme-immunoassays (2). Several manufacturers are now The reproducibility of the assays was studied by evaluatproducing kits for the clinical determination of serum ferritin,
involving various technologies and tracers. These appar suitablefor routine use, but results vary greatly among the different kits (4-9). The ferritins used in preparing immunogens and stan-
dards are extracted from various tissues (spleen, liver, and others) for use in the various kits,and are known to be dissimilar, both structurally and immunologically (10). Ferritin is a heterogeneous protein. The composition and immunological reactivity of isoferritins from various tissues differ and are affected by the physiological state of the cells (11). The variability of determination of serum ferritin with the various kits has been attributed mainly to the immunological heterogeneity of isoferritins (4), the lack of an international standard (5), and the methodology used (12). Consequently, reference intervals and interpretation of the values obtained vary with the kit used (4), making difficult any interlaboratory and other comparisons. Moreover, the immunological reactivity of serum ferritin appears to change during various diseases (13), so that kits that detect different isoferritin types may react differently for healthysubjects as compared with patientswith malignancy and ironoverload(13, 14). Here we report our studies with kits from seven different
manufacturers, intended to clarify the major causes of the variability of serum ferritin evaluation. We used calibrated standards of human liver ferritin (an international reference standard), human heart ferritin, and normal sera. We Ill Servizio Institute
Analisi,
SpedaliCivili, 25100 Brescia, Italy.
of Chemistry,
University
of Brescia,
Via Valsabbina,
25100 Brescia, Italy. 2Department of Biomedical Science and Technology, University
of Milano, Via Olgettina 60, 20131 Milano, Italy. Received July 26, 1983; accepted October 24, 1983. 298
CLINICAL CHEMISTRY, Vol.30, No. 2, 1984
ing the Response Error Relationship (RER) as described by Ekins (15). The standards included with the kits were evaluated 13 times with each of the assays. When the means
of the measures for each standard point were plotted vs the median of the variance (SD2) of the measure of the same point, the slope(RER) of the regression line obtained was an index of assay imprecision. Moreover, the imprecision for each value of the standard curve was calculated from the ratio between the standard deviation of the measure and the slope of the standard curve at that point. The ratio between this value and the concentration
of the standard expresses
the coefficient ofvariation(CV) (15, 16).
Results Evaluation
of Ferritin
Standards
Derived
from Liver
and Heart The concentration of the calibrated NBSB standard, diluted in the kit buffers, was evaluated with the seven assays calibrated with the standard curves provided with the kits. Figure 1 shows the respective regression lines for the relation between the nominal concentration of the reference standard and the results for it with the seven kits. The correlation was highly significant for all the kits (r >0.98);
however, results with five of the kits differed
significantly
from the nominal values. In particular, the Hybritech kit underestimated the concentration by 30% and the Clinical Assays and Hoechst kits overestimated it by 24% and 93%, respectively. To ascertain the cross reactivity of the kits with various
human isoferritins, we measured the concentration of a purified preparation of human heart ferritin that previously had been calibrated on the basis of protein concentration (18). Figure 2 shows the regression lines relating the nominal concentration of this heart ferritin with the results
Table 1. Characteristics of the Kits Source of ferritin
Standard
Method
Manufacturer A. Lepetit B. Clirt. Assays C. Leeco D. Corning E. Hoechst
RIA (DA-SP) RIA (DA-PEG) RIA (DA-PEG) IRMA (GB) IRMA (PB)
Liverc Livers (?) (?) Liver and placenta
LiverU Liver’ (?) Spleen Serum
Incubation
Sample
Antibodyb
Immunogen
vol, L
R R R
200 100 200
R
50
Time, h 1 1
Temp, ‘C 25 37
100
JR
1G
1
37
2.5
25
2
22
10 22 Spleen R Spleen C F. Ramco IRMA (PB) Mouse monoclonal 25 2 IRMA (PB) Liver (?) G. Hybritech Sp solid phase; DA, double antibody; PEG, polyethyleneglycol 6000; GB. glass beads; PB, polystyrene beads; IRMA, immunoradiometric assay. bR,
4 37
rabbit; G, goat.
‘Purified by crystallization with CdSO4.
yO.12,
r1.OQ
yl.?6,
#{149}.15
D
y’O.1S_
r’l.QO
0
yO.2O,
0.98
F
yO.l8.
,0.99
U
yQ.2S.
r’O.99
A B C
B
ri.0O
A,D
100
100
200
NRSB
A
y’l.Ol’
B
y1.24,
C
yl.08,
0
yl.0i,
, , ,
r=U.99
r1.00
500
r1.0O
1000 HUMAN
0
yl.93
,
F
y0.A0
#{149} r0.98
U
y0.?l,
,
STANDARD
C 9/L
r’0.98
1.0O
HEART
FERRITIN
C pg/L
Fig. 2. Measurement of a preparation of human heart ferritin with the sevenkits, all calibrated with the NBSB reference preparation ence standard. Results with the Lepetit and Corning kits, however, were essentially unaffected by this calibration.
Fig.1. NBSB reference preparation and kits’ standards compared The inseit lists the equationsof the best fitted lines and the regression coefficients. The interceptsare omitted because nonewas statistically different fromzero (Studentsf-test:p 0.98) in the range 10-1000 g/L. All kits detected a minor proportion of the heart isoferritins (12-23%) except the
Analytical
Recovery
and Dilution
Tests
For sara with various concentrations of ferritin, tested in serial dilutions, the ferritin concentration as measured by all seven kits decreased linearly with increasing dilution,
the correlation being highly significant (r >0.98, p
